NO/FOMO

Independent AI signal, once a day

The AI briefing worth opening.

ISSUE DATE2026-05-30ENGLISH EDITION
This issue
—
All time
—

Hacker News

3 stories
01

OpenRouter raises $113M Series B

OpenRouter has successfully raised $113 million in a Series B funding round, marking a significant milestone in the developer tools and AI infrastructure landscape. As a prominent aggregator and routing layer for large language models, OpenRouter enables developers to seamlessly access, compare, and deploy a diverse array of open-source and proprietary AI models through a unified API. This substantial capital injection will be used to enhance the platform's infrastructure, optimize latency, and expand its global server footprint to meet the growing demand for flexible model routing. Additionally, the funding will support further development of advanced features such as automated cost optimization, intelligent fallback mechanisms, and custom model hosting, strengthening OpenRouter's position as a critical intermediary in the rapidly evolving generative AI ecosystem.

OpenRouterSeries B FundingModel RoutingUnified APIAI InfrastructureArtificial IntelligenceLarge Language ModelGenerative AI
02

Ernst & Young published cybersecurity report full of hallucinations

Professional services firm Ernst & Young (EY) has reportedly published a cybersecurity report containing significant factual errors and artificial intelligence hallucinations. The investigation reveals that the firm relied on generative AI tools to draft technical cybersecurity assessments, which led to the inclusion of completely fabricated vulnerabilities, non-existent software CVE references, and incorrect threat actor attributions. This incident highlights the critical risks of deploying large language models in professional advisory and security auditing roles without rigorous expert verification. It underscores the ongoing challenges of AI trust, the necessity of human-in-the-loop workflows, and the potential legal and reputational liabilities that corporations face when relying on unchecked AI-generated outputs for highly sensitive technical analyses.

AI HallucinationCybersecurityGenerative AILarge Language ModelsAI AuditingArtificial IntelligenceGenerative AIOthers
03

Leo's first encyclical attacks technological messianism

This article analyzes the newly released papal encyclical by Pope Leo, which delivers a profound critique of technological messianism and the unchecked rise of modern tech-optimism. The encyclical addresses the ethical boundaries of rapid technological advancements, raising serious concerns over the philosophical implications of artificial intelligence, automation, and digital savior complex ideologies. By analyzing how modern societies increasingly rely on technological interventions to solve existential human problems, the Pope warns against replacing traditional moral structures with technocratic solutions. The document advocates for a human-centric approach to future scientific progress, urging world leaders, researchers, and technology developers to align technological innovation with ethical standards, human dignity, and global equity. This high-level philosophical critique serves as a timely reminder of the societal responsibilities inherent in the creation of powerful technological paradigms.

Technological MessianismEthical AITechnological DeterminismHuman-Centric TechnologyEthics of AIArtificial IntelligenceGenerative AIOthers

Twitter

5 stories
01

c_valenzuelab_GenAI Progress

Cristobal Valenzuela, CEO of Runway, shared significant updates regarding the rapid evolution of generative artificial intelligence technologies. The tweet highlights the transformative impact of recent model developments, emphasizing the shift toward high-fidelity video generation and advanced creative tools. By showcasing these cutting-edge capabilities, the message underscores Runway's ongoing commitment to pushing the boundaries of AI-driven cinematic production and content creation. The discussion focuses on the integration of temporal consistency in synthetic media and the practical implications of these tools for professional creative workflows. As the industry moves toward more sophisticated generative models, such insights from key figures reflect the broader movement in the AI sector to merge technical research with high-quality artistic utility, ultimately reshaping the future of digital storytelling and production pipelines in an era of unprecedented technological acceleration.

RunwayGenerative AIVideo GenerationSynthetic MediaCreative ToolsGenerative AIComputer VisionArtificial Intelligence
02

natolambert_Open Science & AI

In this tweet, Nathan Lambert emphasizes the significant role of open science in shaping the discourse surrounding modern artificial intelligence. By highlighting the Tulu 3 project and the introduction of the Reinforcement Learning Verifiable Rewards (RLVR) method, Lambert underscores how public methodologies contribute to standardizing technical communication. He argues that establishing these scientific methods in the public domain provides substantial long-term value, effectively reducing future noise and fragmentation in AI research. This perspective advocates for open transparency as a foundational pillar for setting industry benchmarks and clarifying complex research trajectories. By fostering an environment where protocols are established publicly, researchers can improve scientific rigor and ensure that emerging AI advancements are grounded in accessible, reproducible, and clearly defined technical frameworks that benefit the broader community.

Open ScienceTulu 3RLVRAI ResearchMethodologyArtificial IntelligenceLarge Language ModelOpen Source
03

gdb_Building With Codex

This tweet from user gdb highlights the positive experience of utilizing OpenAI's Codex model for software development tasks. Codex, a descendant of the GPT language models, is specifically fine-tuned on public code from GitHub to assist developers by generating functional code snippets, autocompleting functions, and explaining programming logic. By leveraging the model's advanced natural language understanding and code generation capabilities, developers can significantly accelerate their workflow, reduce boilerplate tasks, and explore new architectural patterns. The statement underscores the ongoing shift toward AI-assisted software engineering, where tools like Codex function as powerful pair programmers, enabling higher productivity and fostering innovation within the developer community. Such professional feedback emphasizes the tangible utility and technical reliability of LLM-based coding assistants in practical, real-world development environments.

CodexOpenAISoftware DevelopmentCode GenerationDeveloper ToolsArtificial IntelligenceLarge Language ModelTech News
04

GaryMarcus_AI Profitability

Gary Marcus, a prominent voice in artificial intelligence discourse, recently shared his analytical perspective regarding the long-term financial viability of leading AI organizations, specifically contrasting Anthropic and OpenAI. In his assessment, Marcus posits that Anthropic possesses a comparatively higher probability of achieving sustainable long-term profitability than its competitor, OpenAI. Despite this comparative advantage, he maintains a cautious stance, suggesting that the likelihood of significant financial success remains relatively low for both entities. This commentary highlights the ongoing industry debate surrounding the business models of frontier AI labs, the immense capital expenditures required to sustain large-scale model development, and the uncertainty regarding monetization strategies within the competitive landscape of generative AI. Marcus's viewpoint reflects broader investor skepticism concerning the current path toward achieving robust, self-sustaining financial growth in the high-stakes AI sector.

AnthropicOpenAIProfitabilityBusiness ModelsAI IndustryArtificial IntelligenceLarge Language ModelIndustry News
05

GaryMarcus_LLM Truth Gaps

Gary Marcus, a prominent critic of current AI trajectories, has reiterated his long-standing stance that Large Language Models (LLMs) suffer from inherent limitations regarding truthfulness and factual accuracy. In his latest commentary, Marcus claims that the industry's reliance on these models remains largely unproductive because they fail to reliably handle objective reality. He defends his historical skepticism against recurring criticism, suggesting that his seven-year record of warnings regarding LLM reliability has been consistently validated by recent developments. This critique underscores a persistent tension between scaling-based AI development and the need for robust reasoning capabilities. The discussion highlights the fundamental debate regarding whether probabilistic token prediction can ever truly achieve the level of truth-grounding necessary for reliable, high-stakes decision-making in real-world applications.

LLMTruthfulnessGary MarcusAI LimitationsReasoningArtificial IntelligenceLarge Language ModelIndustry News